development of a highly protective combination monoclonal antibody therapy against chikungunya virus

Four of 36 neutralizing MAbs CHK-102, CHK-152, CHK-166, and CHK-263 provided complete protection against lethality as prophylaxis in highly susceptible immunocompromised mice lacking the

Monoclonal Antibody Therapy against Chikungunya

Virus

Pankaj Pal1, Kimberly A Dowd2, James D Brien3, Melissa A Edeling4, Sergey Gorlatov5, Syd Johnson5, Iris Lee3, Wataru Akahata6, Gary J Nabel6, Mareike K S Richter7, Jolanda M Smit7, Daved H Fremont4,8, Theodore C Pierson2, Mark T Heise9, Michael S Diamond1,3,8*

1 Department of Molecular Microbiology, Washington University School of Medicine, St Louis, Missouri, United States of America, 2 Viral Pathogenesis Section, Laboratory

Hill, Chapel Hill, North Carolina, United States of America

Abstract

Chikungunya virus (CHIKV) is a mosquito-transmitted alphavirus that causes global epidemics of a debilitating polyarthritis

in humans As there is a pressing need for the development of therapeutic agents, we screened 230 new mouse anti-CHIKV monoclonal antibodies (MAbs) for their ability to inhibit infection of all three CHIKV genotypes Four of 36 neutralizing MAbs (CHK-102, CHK-152, CHK-166, and CHK-263) provided complete protection against lethality as prophylaxis in highly susceptible immunocompromised mice lacking the type I IFN receptor (Ifnar2/2) and mapped to distinct epitopes on the E1 and E2 structural proteins CHK-152, the most protective MAb, was humanized, shown to block viral fusion, and require Fc effector function for optimal activity in vivo In post-exposure therapeutic trials, administration of a single dose of a combination of two neutralizing MAbs (CHK-102+CHK-152 or CHK-166+CHK-152) limited the development of resistance and protected immunocompromised mice against disease when given 24 to 36 hours before CHIKV-induced death Selected pairs of highly neutralizing MAbs may be a promising treatment option for CHIKV in humans

Citation: Pal P, Dowd KA, Brien JD, Edeling MA, Gorlatov S, et al (2013) Development of a Highly Protective Combination Monoclonal Antibody Therapy against Chikungunya Virus PLoS Pathog 9(4): e1003312 doi:10.1371/journal.ppat.1003312

Editor: Fe´lix A Rey, Institut Pasteur, France

Received November 20, 2012; Accepted March 4, 2013; Published April 18, 2013

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose The work is made available under the Creative Commons CC0 public domain dedication.

Funding: U54 AI 057157: Southeast Regional Center of Excellence for Emerging Infections and Biodefense (to MTH), NIH grant R01-AI104545, and the Intramural Research Program of the National Institutes of Allergy and Infectious Diseases (NIAID) supported this work The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests: I have read the journal’s policy and have the following conflicts: S Johnson and S Gorlatov are employees of MacroGenics, which has an option for licensure of the anti-CHKV MAbs for commercial development M Diamond is a paid consultant for MacroGenics This does not alter our adherence to all PLoS Pathogens policies on sharing data and materials.

* E-mail: diamond@borcim.wustl.edu

Introduction

Chikungunya virus (CHIKV) infection causes a severe febrile

illness in humans that is characterized by a debilitating

polyar-thritis, which can persist for months and cause significant

morbidity [1,2] There are three genotypes of CHIKV: Asian,

East/Central/South African (ECSA), and West African [3–5],

with 95.2 to 99.8% amino acid identity [4] The CHIKV strains

from the recent epidemics belong to the ECSA genotype and have

affected millions in Africa and the Indian subcontinent [3,6]

Imported cases in the United States and outbreaks in Europe

highlight the threat of CHIKV to developed countries [7]

Currently, there are no approved vaccines or therapeutics for

CHIKV [8]

CHIKV is an enveloped alphavirus of the Togaviridae family that

enters cells via receptor-mediated internalization and a low

pH-triggered type II membrane fusion event in early endosomes The

mature virion is comprised of three structural proteins: a nucleocapsid protein and two glycoproteins, E1 and E2, where E2 functions in attachment to cells and E1 participates in virus fusion Each 700 A˚ CHIKV virion contains 240 copies of the envelope and capsid proteins, which are arranged in T = 4 quasi-icosahedral symmetry E1-E2 heterodimers assemble into 80 trimeric spikes on the virus surface [9] X-ray crystallographic structures of the precursor pE3-E2-E1, mature E2-E1, and E1 proteins [10–13] have elucidated the architecture of the glycopro-tein shell The E1 ectodomain consists of three domains Domain I (DI) is located between DII and DIII, the latter of which adopts an immunoglobulin-like fold The fusion peptide is located at the distal end of DII E1 monomers lie at the base of the surface spikes and form a trimer around each of the icosahedral axes E2 localizes to a long, thin leaf-like structure on the top of the spike The mature E2 protein contains three domains with immuno-globulin-like folds: the N-terminal domain A, located at the center;

domain B at the tip; and the C-terminal domain C, located

proximal to the viral membrane

Mouse models have been developed for CHIKV infection

Newborn outbred and inbred mice are vulnerable to severe

CHIKV infection with viral replication observed in muscle, joint,

and skin [14,15] Adult mice with defects in type I interferon

signaling (Ifnar2/2mice) develop lethal disease, with muscle, joint,

and skin appearing as the primary sites of infection [15] CHIKV

infection of juvenile C57BL/6 mice by a subcutaneous route

results in metatarsal foot swelling with histological evidence of

arthritis, tenosynovitis and myositis [16,17]

Passive transfer of MAbs or immune sera can protect animals

against infection of alphaviruses including Sindbis (SINV), Semliki

Forest (SFV), and Venezuelan equine encephalitis (VEEV) viruses

[18–25] Immune c-globulin from human donors in the

conva-lescent phase of CHIKV infection exhibited neutralizing activity in

vitro and had partial therapeutic efficacy in Ifnar2/2and neonatal

wild type mice when administered up to 24 hours after infection

[26] Although mouse and human MAbs that neutralize CHIKV

infection have been reported [27,28], their post-exposure efficacy

against lethal infection in vivo has not been clearly established [29]

Here, we investigated the molecular basis of antibody-mediated

neutralization of CHIKV using a panel of 230 newly generated,

cloned MAbs CHK-152 protected mice against CHIKV-induced

mortality and disease The inclusion of a second MAb (CHK-166

or CHK-102) prevented the emergence of viral resistance and

extended the treatment window in Ifnar2/2 mice up to 24 to

36 hours prior to death of the animals Our results suggest that

combination therapy with selected neutralizing MAbs has

potential for treatment of CHIKV infection in humans

Results

Generation of MAbs

We generated a panel of neutralizing MAbs against CHIKV as

a first step towards a possible therapy in humans We infected

adult C57BL/6 mice deficient for interferon regulatory factor 7

(Irf72/2) with 104PFU of the La Reunion 2006 OPY-1 strain of

CHIKV (CHIKV-LR); these mice were boosted with CHIK virus-like particles [30], soluble recombinant CHIKV E2 protein, or live CHIKV-LR We immunized Irf72/2rather than wild type (WT) mice, as CHIKV replicated to higher titers, induced stronger neutralizing antibody responses, yet did not cause lethal infection

in these innate immune-deficient animals ([31], and data not shown) We screened four independent myeloma cell-splenocyte fusions for binding of hybridoma supernatants to CHIKV-LR infected cells (Fig S1) and cloned 230 CHIKV-specific MAbs for further analysis (Table S1 in Text S1) Using a single endpoint neutralization assay, we identified 36 MAbs with inhibitory activity against infection of CHIKV-LR in BHK21-15 cells (data not shown)

Neutralizing activity

To assess the inhibitory potential of our anti-CHIKV MAbs against the homologous CHIKV-LR and representative strains from the Asian and West African genotypes (RSU1 and IbH35 respectively), we performed focus reduction neutralization tests (FRNTs) on Vero cells We determined the concentration of MAb that reduced the number of foci of infection by 50 or 90% (EC50 and EC90 values,Fig 1A and B, and Table 1) CHK-152 was the most strongly neutralizing MAb we identified; 3 and 15 ng/ml

of this MAb prevented 50 and 90% of CHIKV infection against all three CHIKV genotypes (Fig 1C) Ten other MAbs inhibited CHIKV infection with EC50 values of ,10 ng/ml against all three genotypes, and many others inhibited all three strains similarly, with a few exceptions For example, CHK-9 failed to neutralize the Asian strain to the same extent as the West African

or La Reunion (ECSA genotype) strains (Fig 1D), whereas

CHK-151 inhibited infection of the Asian strain better than the others (Table 1) Also, for reasons that are unclear, some neutralizing MAbs (e.g., CHK-143, CHK-264, and CHK-269) were incapable

of inhibiting all viruses (EC90.10,000 ng/ml) in this assay, even

at high MAb concentrations

We speculated that some MAbs might show cell type-dependent neutralization if they blocked attachment to cell type-specific factors To test this hypothesis, we assessed MAb neutralization of CHIKV-LR infection in cells of another species, NIH 3T3 mouse fibroblasts (Table 1) For most MAbs, the EC50 values were comparable to those achieved with Vero cells However, two MAbs (CHK-96 and CHK-176) showed a 12 to 250-fold reduction (P,0.05) in neutralizing activity on NIH 3T3 compared

to Vero cells; although further study is warranted, these MAbs may block a step in the entry pathway that varies among different cell types

Prophylaxis studies

To evaluate whether neutralizing MAbs protect against CHIKV infection in vivo, we initially used a stringent test model: prevention of lethal infection in immunodeficient Ifnar2/2 C57BL/6 mice One hundred micrograms of 14 different MAbs with strong, modest, or poor neutralizing activity were adminis-tered to Ifnar2/2mice one day prior to CHIKV-LR infection As seen previously [15], all Ifnar2/2mice died by day 4 after infection when treated with saline or a negative control MAb (Fig 2A, and data not shown) Strongly neutralizing (e.g., CHK-102, CHK-152, and CHK-263) and one moderately inhibitory (CHK-166) MAb protected 100% of mice from lethal infection (P,0.0001) In comparison, and somewhat surprisingly, CHK-95, a potently neutralizing MAb of the same IgG2c isotype, protected only 12%

of mice from death The other MAbs tested conferred interme-diate levels of protection (Fig 2A) Thus, although several strongly neutralizing MAbs prevented against lethal CHIKV infection in

Author Summary

Chikungunya virus (CHIKV) is a mosquito-transmitted

alphavirus that causes outbreaks of polyarthritis in

humans, and is currently a threat to spread to the United

States due to the presence of its mosquito vector, Aedes

albopictus At present, there is no licensed human vaccine

or therapeutic available to protect against CHIKV infection

The primary goal of this study was to develop an

antibody-based therapeutic agent against CHIKV To do this, we

developed a panel of 230 new mouse anti-CHIKV MAbs

and tested them for their ability to neutralize infection of

different CHIKV strains in cell culture We identified 36

MAbs with broad neutralizing activity, and then tested

several of these for their ability to protect

immunocom-promised Ifnar2/2mice against lethal CHIKV infection In

post-exposure therapeutic trials, administration of a single

dose of a combination of two neutralizing MAbs limited

the development of resistance and protected Ifnar2/2

mice against disease even when given just 24 to 36 hours

before CHIKV-induced death Analogous protection

against CHIKV-induced arthritis was seen in a disease

model in wild type mice Our data suggest that pairs of

highly neutralizing MAbs may be a therapeutic option

against CHIKV infection

Ifnar2/2mice, in vitro neutralization activity per se did not directly

correlate with protection To define the relative potency of the

four MAbs that completely prevented lethal disease, we

admin-istered a lower (10mg) dose Whereas CHK-152 and CHK-263

still protected most mice from lethal infection, CHK-102 and

CHK-166 protected to a lesser degree or only prolonged survival

(Fig 2B) Consistent with their ability to protect against lethal

infection, passive transfer of CHK-102, CHK-152, CHK-166, and

CHK-263 MAbs all markedly reduced viral loads in serum,

spleen, liver, muscle, and brain at 48 hours after infection relative

to a non-binding isotype control (DENV1-E98) MAb (Fig 2C–

G) The level of protection afforded by CHK-102, CHK-152,

CHK-166, and CHK-263 MAbs, however, did not correlate

directly with their binding strength to CHIKV surface

glycopro-teins (Fig S2)

Although a stringent test of MAb protection, CHIKV-infected

Ifnar2/2mice do not develop the arthritis observed in humans To

evaluate this, we utilized a WT C57BL/6 mouse model in which

inoculation of CHIKV into the footpad results in localized swelling

and induction of arthritis and fasciitis within the foot and ankle

[16,17], although infection does not cause lethality Pretreatment

of mice with either 100mg of CHK-102 or CHK-152 completely protected against CHIKV-induced swelling, compared to control animals, which developed clinically apparent swelling (data not shown) While CHIKV infected control animals developed inflammatory arthritis in the ankle and foot, CHK-102 or CHK-152 MAb treated animals had normal appearing joint tissues (Fig 2H)

Mechanism of neutralization

Antibody neutralization of enveloped viruses can occur by inhibiting attachment, internalization, and/or fusion [32,33] To determine how many of our most protective MAbs inhibited infection in cell culture, we performed pre- and post-attachment neutralization assays [34,35] Anti-CHK MAbs were incubated with CHIKV before or after virus binding to cells, and infection was measured As expected, all MAbs efficiently neutralized infection when pre-mixed with virus (Fig 3A) While CHK-102, CHK-152, CHK-166, and CHK-263 also inhibited CHIKV infection when added after virus adsorption to the cell surface,

Figure 1 Profile of neutralizing MAbs against CHIKV A Examples of MAb neutralization as judged by a reduction in the number of FFU using the Biospot Macroanalyzer Rows 2 to 12 going across represent decreasing (3-fold) concentrations of CHK-152 or the negative control DENV1-E98 MAb Column 1 shows infection in the absence of MAb B Increasing concentrations of CHK-95, CHK-102, CHK-166, CHK-187, or CHK-263 were mixed with 100 to 150 FFU of CHIKV-LR for one hour at 37uC and Vero cells were infected Neutralization was determined by FFU assay C–D CHK-152 (C) or CHK-9 (D) was mixed with CHIKV-LR (East, Central and South African genotype), CHIKV-RSUI (Asian genotype), or CHIKV IbH35 (West African genotype) for one hour at 37uC and Vero or NIH 3T3 cells were infected as indicated Neutralization was determined by FFU assay Data in this Figure

is pooled from three independent experiments performed in duplicate or triplicate All error bars represent the standard deviations.

doi:10.1371/journal.ppat.1003312.g001

Table

suggesting that at least part of their blocking activity was at a

post-attachment step, differences in the extent of neutralization were

noted in this context for several MAbs CHK-152 completely

neutralized all CHIKV virions without a resistant fraction when

added post-attachment When studies were repeated with eight

other neutralizing MAbs that showed pre-exposure protection in

vivo, no other MAb inhibited infection completely when added

after virus adsorption to the cell As expected, an isotype control

MAb (DENV1-E98) and a non-neutralizing anti-CHK MAb

(CHK-84) had no inhibitory effects in this assay (Fig S3)

Blockade of viral fusion

Since CHK-152 neutralized infection efficiently at a post-attachment step, we investigated whether it blocked fusion using a viral fusion from without (FFWO) assay [36] CHIKV was adsorbed to Vero cell monolayers on ice and then treated with MAbs Fusion at the plasma membrane was triggered after a brief exposure to low pH buffered medium at 37uC Subsequently, cells were incubated in the presence of 20 mM NH4Cl to prevent CHIKV fusion via canonical endosomal pathways As expected, at

14 hours after initial treatment, CHIKV infection was not

Figure 2 Efficacy of anti-CHIKV MAb prophylaxis A Six to eight week-old Ifnar 2/2 C57BL/6 mice were passively transferred 100 mg of the indicated MAbs via an i.p injection one day before infection with 10 FFU of CHIKV-LR via a s.c route The percentage and number of surviving mice were as follows: DENV1-E98 (0%, 0 of 9), CHK-88 (62.5%; 5 of 8), CHK-95 (12.5%; 1 of 8), CHK-98 (28.6%; 2 of 7), CHK-102 (100%; 8 of 8), CHK-124 (75%;

6 of 8), CHK-151 (87.5%; 7 of 8), CHK-152 (100%; 8 of 8), CHK-155 (85.7%; 6 of 7), CHK-165 (28.6%; 2 of 7), CHK-166 (100%; 8 of 8), CHK-175 (75%; 6 of 8), CHK-187 (50%; 4 of 8), CHK-263 (100%; 8 of 8), or CHK-266 (0%; 0 of 8) MAbs italicized in red in the Figure provided 100% protection B Ifnar 2/2

mice were passively transferred 10 mg of MAb via an i.p injection one day before infection with 10 FFU of CHIKV-LR via a s.c route The percentage and number of surviving mice were as follows: DENV1-E98 (0%; 0 of 7), CHK-102 (12.5%; 1 of 8), CHK-152 (83%; 10 of 12), CHK-166 (0%; 0 of 12), or CHK-263 (73%; 8 of 11) For (A) and (B) the survival curves were constructed from data of at least two independent experiments All anti-CHK MAbs provided statistically significant protection in the percentage of surviving animals or mean survival time compared to the control DENV1-E98 MAb (P,0.05) C–G Viral burden in MAb-treated Ifnar 2/2

mice Animals were passively transferred 100 mg of the indicated MAbs (102, 152,

CHK-166, CHK-263, or isotype control DENV1-E98) via an i.p injection one day before infection with 10 FFU of CHIKV-LR via a s.c route Two days later, viremia (C) and tissues (D, spleen; E, liver; F, muscle; and G, brain) were harvested and infectious virus was titrated by focus-forming assay Results are pooled from two independent experiments (n = 4 mice per group) The dashed line indicates the limit of detection of the assay and the solid bar indicates the median values All viral burden results with CHK-102, CHK-152, CHK-166, and CHK-263 were statistically different (P,0.02) from those obtained with DENV1-E98, as analyzed by the Mann-Whitney test H Four week-old female WT C57BL/6 mice were sham-treated or administered

100 mg of CHK-102 or CHK-152 via an i.p route 24 hours later, mice were infected with 100 PFU of CHIKV-SL 15649 and at day 10, virus-induced pathology in the foot and ankle joint was assessed (Outer left) Sham-infected, (middle left) CHIKV infected and sham-treated, (middle right) CHIKV-infected and CHK-102 treated, and (outer right) CHIKV CHIKV-infected and CHK-152 treated Shown are representative images after hematoxylin and eosin staining from at least 3 mice per group at 1006 magnification Yellow and green arrows indicate regions of inflammation or normal joints, respectively.

doi:10.1371/journal.ppat.1003312.g002

Figure 3 Mechanism of neutralization by CHIKV MAbs A Pre- and post-attachment inhibition assays Vero cells were pre-chilled to 4uC and

100 FFU of CHIKV-LR was added to each well for one hour After extensive washing at 4uC, the indicated MAbs were added for one hour at 4uC, and then the FRNT protocol was completed (black lines, Post) In comparison, a standard pre-incubation FRNT with all steps performed at 4uC is shown for reference Virus and MAb are incubated together for one hour at 4uC, prior to addition to cells (red lines, Pre) Data shown are representative of three experiments performed in duplicate with error bars representing standard deviation B–C FFWO assay CHIKV was incubated with Vero cells at 4uC to allow virus attachment Free virus was removed after washing and 50 mg/ml of the indicated MAbs (including DENV1-E98, a negative control MAb) were added at 4uC Viral fusion at the plasma membrane was induced after a brief exposure to a low pH buffer After pH normalization, cells were cultured for 14 hours in the presence of NH 4 Cl to inhibit infection through the endosomal pathway Cells were analyzed for infection by staining with

an anti-E2 MAb Representative histograms are shown (B) and the data was pooled from four independent experiments for statistical analysis (C) For simplicity of display, not all of the MAbs included in the summary graph are shown by flow cytometry analysis Asterisks indicate values that are statistically different (P,0.05) from the control MAb Error bars represent standard deviations Note low pH-triggered viral fusion at the plasma membrane is an inefficient process with only 10 to 20% of cells becoming infected even when a high MOI was used D–E Viral membrane fusion with liposomes Fusion of pyrene-labeled CHIKV was measured at pH 4.7 (37uC) using liposomes consisting of phosphatidylcholine, phosphatidyleth-anolamine, sphingomyelin, and cholesterol in a molar ratio of (1/1/1/1.5), as described in the Methods (D) Curve a, no MAb; curve b, 0.1 nM CHK-152; curve c, 1 nM CHK-152; curve d, 10 nM CHK-152 (E) Extent of fusion (average value between 50 to 60 seconds post acidification) at increasing concentrations of MAb Black bars, CHK-152; white bar, isotype control (MAb 0031, only included at 10 nM concentration) All fusion measurements were performed at least three independent times.

doi:10.1371/journal.ppat.1003312.g003

observed when adsorbed virus was incubated at neutral pH

(Fig 3B) In comparison, in the absence of MAb or in the

presence of a control MAb, a short exposure of cell

surface-adsorbed virus to acidic pH resulted in infection and

CHIKV-antigen positive cells Notably, CHK-152 completely inhibited

(P,0.0001) plasma membrane fusion and infection, whereas other

anti-CHIKV neutralizing MAbs showed significant yet incomplete

inhibition in this assay (Fig 3B and C) These studies suggest that

CHK-152 efficiently neutralizes infection by preventing the

structural changes on the virion necessary for viral fusion with

host cell membranes

We utilized a model liposome fusion assay with pyrene-labeled

virus [37,38] to confirm these results Pyrene-labeled CHIKV was

pre-incubated with different concentrations of MAb, mixed with

liposomes at 37uC, and fusion was triggered by addition of a

low-pH buffer [37] In the absence of MAb or in the presence of

10 nM (1.5mg/ml) of a non-binding control MAb, fusion was

complete within seconds of acidification In contrast,

pre-incubation of virus with increasing doses of CHK-152 inhibited

fusion (Fig 3D and E) Thus, CHK-152 can block

low-pH-induced fusion of virus with liposomes

The effector functions of CHK-152 contribute to

protection in vivo

To define additional mechanisms by which our most strongly

protective MAb (CHK-152) conferred protection in vivo, we

generated a chimeric mouse-human CHK-152 (ch-CHK-152) as

well as an aglycosyl variant (ch-CHK-152 N297Q) that lacks the

ability to engage C1q or Fc-c receptors; this mutation does not

affect the ability to bind the neonatal Fc receptor (FcRn) or

half-life of antibody in mouse serum [39] The affinity of ch-CHK-152

and ch-CHK-152 N297Q binding to purified pE2-E1 was

measured by surface plasmon resonance (SPR) and compared to

the parent murine MAb Notably, ch-CHK-152, ch-CHK-152

N297Q, and the murine CHK-152 all had similar affinity (KDof 3

to 4 nM) (Fig 4A and data not shown) and neutralizing activity in

cell culture (Fig 4B) As expected, ch-CHK-152 N297Q failed to

bind efficiently to soluble Fc-c receptors or C1q (Fig 4C)

We transferred ch-CHK-152 and ch-CHK152 N297Q to

Ifnar2/2mice prior to infection Although high doses (100mg) of

ch-CHK-152 and ch-CHK-152 N297Q provided similar

protec-tion against CHIKV infecprotec-tion (data not shown), lower doses

(10mg) of the aglycosyl variant were less protective; whereas 62%

of the mice receiving ch-CHK152 N297Q survived, all Ifnar2/2

mice given ch-CHK-152 MAb remained alive (Fig 4D, P,0.05)

When parallel studies were performed with WT C57BL/6 mice

and MAb was administered 18 hours after infection, ch-CHK-152

N297Q also provided less protection against arthritis compared to

ch-CHK-152 (Fig 4E) These data suggest that the Fc effector

interactions contribute to the potency of CHK-152 in mice

Humanization of CHK-152

We humanized CHK-152 as a first step towards a MAb

therapeutic (seeText S1) The affinity for pE2-E1 and

neutral-izing activity of the hu-CHK-152 were similar to mouse CHK-152

(Fig S4A and B) Hu-CHK-152 also protected Ifnar2/2 mice

(P.0.0001) when a single dose (10 or 100mg) was administered

one day before infection (Fig S4C)

Therapeutic studies

To define the therapeutic potential of our most protective

MAbs, a single dose (100mg) was administered to Ifnar2/2mice

24 hours after CHIKV infection (Fig 5A) Whereas CHK-152

and 166 protected 58% and 63% of mice from death, respectively (P,0.0001), CHK-263 and CHK-102 had less activity although both MAbs increased the median survival time (7 days versus 4 days with the control DENV1-E98 MAb, P,0.0006) Adminis-tration of CHK-152 at 12 or 18 hours post infection also protected

WT mice from CHIKV-induced swelling and arthritis (Fig 5B and Fig 4E)

We next tested the activity of combinations of the most protective neutralizing MAbs in Ifnar2/2 mice Remarkably, administration of 50mg each (100mg total dose) of CHK-102+CHK-152, CHK-263+CHK-152, or CHK-166+CHK-152

at 24 hours post infection completely prevented mortality in all animals (Fig 5A, P,0.0001 for MAb combinations) This observation was not true for all MAb combinations, as adminis-tration of 50mg each of CHK-102+CHK-263 provided substan-tially less protection with a 14% survival rate We then performed

a more stringent test in which 100mg each (200mg total) of our most protective combinations was delivered as a single dose at

48 hours post-infection (Fig 5C) Treatment with CHK-102+CHK-152 or CHK-166+CHK-152 protected 62% of the Ifnar2/2 mice (P,0.003) and the combination of CHK-263+CHK-152 functioned almost as well, with 50% of animals surviving (P,0.03) To define the limits of protection in Ifnar2/2 mice, which all succumb to CHIKV between days 3 and 4, therapy was initiated at 60 and 72 hours after infection At

60 hours after infection, Ifnar2/2mice receiving 250mg each of CHK-102+CHK-152 or CHK-166+CHK-152 had survival rates

of 28 and 71%, respectively (Fig 5D, P = 0.03 and P = 0.004) Nonetheless, when combination therapy was given at 72 hours after infection, a time when overt disease was present, no survival benefit was conferred Thus, combination MAb therapy is superior

to monotherapy in protecting against lethal CHIKV infection in highly immunocompromised mice

Functional interaction of MAbs

To begin to understand the basis for enhanced in vivo activity,

we assessed whether CHK-152 and selected MAbs could bind simultaneously to the CHIKV virion We developed a competition ELISA in which virions were captured by a mouse MAb (65), and then incubated with increasing concentrations of

CHK-102, CHK-152, CHK-166, or CHK-263 mouse MAbs After washing, hu-CHK-152 MAb was added, and binding was assessed While pre-bound mouse CHK-152 competed against hu-CHK-152 binding as expected, CHK-102, CHK-166, and CHK-263 minimally competed hu-CHK-152 binding (Fig S5A), suggesting their epitopes largely were distinct However, addition

of CHK-102, CHK-166, or CHK-263 failed to augment the inhibitory activity of CHK-152 when neutralization was measured

in cell culture (Fig S5B), as no synergy was observed

Neutralization escape mutants

To identify epitopes targeted by the therapeutic MAbs, we generated escape mutants in cell culture After sequential virus passage under CHK-102, CHK-152, CHK-166, or CHK-263 selection, CHIKV became resistant to neutralization by these MAbs (Fig 6A–D) We assessed whether the escape variants generated in the presence of one MAb remained sensitive to neutralization by the other MAbs The CHK-152 escape variant was neutralized efficiently by 102, 166, and

CHK-263 (Fig 6B, Table S2 in Text S1, and data not shown), and analogously the 166 escape variant was inhibited by

CHK-102, CHK-152, and CHK-263 (Fig 6C, and data not shown) In contrast, CHK-102 and CHK-263 escape variants reciprocally were resistant, suggesting their epitopes were the same or

overlapping (Fig 6A and D); however, CHK-102 and CHK-263

escape variants remained sensitive to neutralization by CHK-152

and CHK-166 Notably, selection with combinations of MAbs

(e.g., CHK-102+CHK-152) failed to produce escape variants

despite several independent attempts (data not shown)

To identify the mutations that conferred resistance, we

sequenced plaque-purified escape variants (Table 2, top) Six of

eight sequences from CHK-102 escape variants contained an L210P mutation in the E2 protein; the remaining two sequences had a G209E mutation in E2 For CHK-152 resistant variants, all sequences (9 of 9) contained a D59N mutation in E2 and two contained a second A89E substitution

in E2 For CHK-263, 3 of 4 escape variants had a K215E change in E2, whereas 1 of 4 had mutations in E2 at G209E

Figure 4 The effector functions of CHK-152 contribute to protectionin vivo A Comparison of binding of CHK-152 and agylocsyl ch-CHK-152 N297Q to pE2-E1, as measured by surface plasmon resonance A single representative sensogram is shown for each MAb The experimental curves (colored lines) were fit using a 1:1 Langmuir analysis (dashed lines), after double referencing, to determine the kinetic parameters presented in the Table immediately below B Comparison of neutralizing activity of murine CHK-152, ch-CHK-152, and ch-CHK-152 N297Q, as measured by FRNT

on Vero cells C Comparison of binding of ch-CHK-152 and ch-CHK-152 N297Q to FccR (CD16A, 500 nM; CD32A, 100 nM; and CD64, 100 nM) or C1q (50 nM), as measured by surface plasmon resonance D Comparison of pre-exposure protective activity of ch-CHK-152 and ch-CHK-152 N297Q Ifnar2/2mice were administered via an i.p injection 10 mg of ch-CHK-152 and ch-CHK-152 N297Q one day before infection with 10 FFU of CHIKV-LR via a s.c route Mice were monitored for survival for 21 days after infection The survival curves were constructed from data of at least two independent experiments and the number of animals for each antibody ranged from 8 to 10 per group ch-CHK-152 provided statistically greater protection than ch-CHK-152 N297Q (P,0.05) E Five week-old WT C57BL/6 mice were infected with 100 PFU of CHIKV in the left rear footpad and either sham-treated, or treated with 100, 50, or 25 mg of ch-CHK-152 (left panel) or ch-CHK-152 N297Q (right panel) at 18 hours post infection Mice were scored daily for virus-induced footpad swelling, where a score of 0 = no swelling, 1 = mild swelling where the top of the foot is slightly raised,

2 = moderate swelling with the entire top of foot raised, and 3 = severe swelling involving both the top and bottom of the foot Scores are the mean values for 7 to 8 mice per treatment group and are representative of three independent experiments Ch-CHK-152 mediated protection was significantly greater than ch-CHK-152 N297Q on days 7, 8, and 9 post infection for the 100 mg antibody dose, and at day 7 post infection for the 50 mg dose, as determined by the Kruskal-Wallace test with Bonferroni correction (P,0.05) No statistically significant differences between ch-CHK-152 and ch-CHK-152 N297Q were observed with the 25 mg dose Of note, we observed a reproducible decrease in clinical score on day 5 in many animals This reflects the biphasic pattern of swelling: during the first 3 to 4 days, swelling is due to edema, whereas after day 5, it is due to inflammatory cell infiltration into the foot and ankle.

doi:10.1371/journal.ppat.1003312.g004

All escape variants (14 of 14) of CHK-166 had a single K61T

mutation in the E1 protein

To verify the amino acid changes that conferred MAb resistance

in vitro, we introduced several of these substitutions into a chimeric

SFV-GFP-CHIKV cDNA comprised of SFV non-structural genes,

a GFP reporter gene, and the CHIKV structural genes (T Lin, K

Dowd, and T Pierson, unpublished results) Parental and

SFV-GFP-CHIKV with single amino acid mutations were analyzed for

neutralization by 102, 152, 166, and

CHK-263 (Fig 6E–H) Consistent with our sequencing results, viruses encoding mutations in E2-G209 and E2-L210 were resistant to CHK-102, changes in E2-D59 conferred resistance to CHK-152, substitutions in E1-K61 resulted in resistance to CHK-166, and mutation of E2-G209 and E2-K215 caused resistance to

CHK-263 However, introduction of E2-A89E (which was present in 2 of

9 clones) failed to affect the neutralizing activity of CHK-152

In addition to selecting escape variants in cell culture, we harvested organs from the few mice that became ill after infection

Figure 5 Therapeutic efficacy of anti-CHIKV MAbs A Ifnar 2/2 mice were passively transferred via an i.p injection 100 mg of DENV1-E98,

CHK-102, CHK-152, CHK-166, or CHK-263 or 50 mg each of CHK-102+CHK-152, CHK-166+CHK-152, CHK-263+CHK-152, or CHK-102+CHK-263 at 24 hours after CHIKV infection B Five week-old WT C57BL/6 mice were infected with 100 PFU of CHIKV in the footpad and either sham-treated, or treated with

100 or 50 mg of CHK-152 at 18 hours post infection Virus induced pathology in the foot and ankle joint was assessed by histopathological analysis at day 10 post-infection (Left) CHIKV-infected, sham-treated, (middle) CHIKV-infected, CHK-152 (100 mg) treated at +18 hours, and (right) CHIKV-infected, CHK-152 (50 mg) treated at +18 hours Shown are representative images after hematoxylin and eosin staining from 3 mice per group at 1006 magnification Yellow and green arrows indicate regions of inflammation or normal joints, respectively C Ifnar2/2mice were passively transferred via

an i.p injection 200 mg of DENV1-E98 or 100 mg each of CHK-102+CHK-152, CHK-166+CHK-152, or CHK-263+CHK-152 at 48 hours after CHIKV infection D Ifnar2/2mice were passively transferred via an i.p injection 500 mg of DENV1-E98 or 250 mg each of 102+152 or CHK-166+CHK-152 at 60 hours after CHIKV infection For A, C, and D the survival curves were constructed from data of at least two independent experiments The number of animals for each antibody ranged from 8 to 10 per group, with the exception of CHK-102+CHK-263, which was performed with 7 mice only Statistically significant differences in protection compared to DENV1-E98 are described in the text.

doi:10.1371/journal.ppat.1003312.g005

Figure 6 Characterization and mapping of neutralization escape mutants A–D FRNT assay with bulk virus obtained after three to six passages under selection of (A) CHK-102, (B) CHK-152, (C) CHK-166, or (D) CHK-263 on Vero cells Bulk virus also was tested for infectivity in the presence of the non-selecting MAbs Results are representative of two to three independent experiments performed in triplicate E–H Confirmation

of resistant phenotype with SFV-CHIKV-GFP containing the indicated single engineered point mutations Serial dilutions of MAb were incubated with chimeric SFV-CHIKV virus (WT or mutant stocks) for one hour at room temperature MAb-virus complexes were added to Vero cells plated in 96-well plates and incubated at 37uC After 8 hours cells were trypsinized, fixed, and the number of GFP-positive, infected cells was assessed by flow cytometry Curves are representative of 2 to 3 independent experiments I Epitope mapping of anti-CHIKV MAbs on the crystal structure of the mature envelope glycoprotein complex (PDB code 3N44) (Left) The domains on E2 (cyan) and E1 (gold) are indicated, and the fusion loop on E1 (E1 FL) is delineated Amino acid residues of neutralizing MAbs were determined by escape selection, sequencing, and reverse genetic confirmation CHK-102 and CHK-263 recognize the B domain on E2, CHK-152 recognizes a residue on the wings of the A domain on E2, and CHK-166 recognizes an amino acid in domain II of E1 proximal to the conserved fusion loop (Right) The mature envelope glycoprotein docked onto the trimer conformation (PDB code 2XFB) that is present on the virion E3, E2, and E1 and the escape residues are colored as in the left panel Neutralization escape residues are readily accessible on the top of the trimer, distal to the viral membrane.

doi:10.1371/journal.ppat.1003312.g006